> also switches back to > the HDF5 serial variant because we cannot provide only the MPI variant.
Dang, this was the most important thing for me. I have been under the assumption that using MPI provides all the capabilities that the serial variant provides, plus parallel computation. Were build failures related to the fact that we used to ship MPI versions? Cheers, Nico On Sat, Aug 1, 2015 at 2:43 AM Sebastiaan Couwenberg <[email protected]> wrote: > On 31-07-15 00:15, Sebastiaan Couwenberg wrote: > > On 30-07-15 14:03, Sebastiaan Couwenberg wrote: > >> There are a couple of easy to solve FTBFS that cannot find the hdf5 > >> library that will need a patch add the HDF5 variant specific library > >> path. For those issues bugs haven't been filed yet, for all other bugs > >> related to the netcdf transition see: > >> > >> > https://bugs.debian.org/cgi-bin/pkgreport.cgi?tag=netcdf-split-c-f-cxx;[email protected] > > > > To address the hdf5 linking issues I've patched nc-config & netcdf.pc to > > include the HDF5 library & include paths as mentiond in the 'HDF5 > > linking issues with new NetCDF packages' thread [1]. These changes will > > be included in netcdf (1:4.4.0~rc2-1~exp4) which I haven't uploaded yet, > > but only use for my rebuilds at the moment. > > netcdf (1:4.4.0~rc2-1~exp4) has been uploaded to experimental, it > includes the nc-config & netcdf.pc changes, and also switches back to > the HDF5 serial variant because we cannot provide only the MPI variant. > Providing both is a bit too much work to do properly before the GCC 5 > related transitions start. > > To not require the build dependency changes for the netcdf-cxx-legacy > and netcdf-fortran -dev packages, we should consider having > libnetcdf-dev recommend the language specific -dev packages. > > I've started a new round of rebuilds with the updated netcdf packages, > and it looks much better than the first round. > > adios (1.8.0-4~exp1) was uploaded to unstable recently, it includes the > build-dependency on libnetcdff-dev (netcdf-fortran) which is not > available in unstable yet. The buildds cannot build it, so the build > dependency needs to be removed in unstable. It reported this issue in > #794230. > > ncview (2.1.5+ds1-1) still FTBFS because it didn't consider the > /usr/bin/cc symlink to gcc acceptable. I've added a patch in > ncview (2.1.5+ds1-2) to disable these unhelpfull configure checks. > > I didn't manage to finish the round of rebuild before GCC 5 was made the > default compiler in unstable causing dependency installation issues. > Except gmt (5.1.2+dfsg1-1 / 5.1.2+dfsg1-2~exp1) all remaining packages > built successfully in the last round, so I expect them to still do with > the updated netcdf packages. I also suspect the gmt FTBFS is fixed now > without needing the changes in experimental. > > While I'd liked to have finished my last round of rebuilds, I'd think > we're ready for the netcdf transition. > > > Transition: netcdf > > libnetcdfc7 (1:4.1.3-7.2) -> libnetcdf7 (1:4.4.0~rc2-1~exp4) > libnetcdfc++4 (1:4.1.3-7.2) -> libnetcdf-c++4-1 (4.2.1-1~exp4) [cxx4] > libnetcdfc++4 (1:4.1.3-7.2) -> libnetcdf-c++4 (4.2-1~exp4) [legacy] > libnetcdff5 (1:4.1.3-7.2) -> libnetcdff6 (4.4.2-1~exp5) > > The status of the most recent rebuilds is as follows. Entries tagged > with [+] build successfully after applying the patch from the bugreport. > > netcdf-cxx (4.2.1-1~exp4) OK > netcdf-cxx-legacy (4.2-1~exp4) OK > netcdf-fortran (4.4.2-1~exp5) OK > > adios (1.8.0-4~exp1) OK [+] > cdftools (3.0-1 / 3.0-2~exp1) FTBFS / OK > cmor (2.9.1-5 / 2.9.1-6) OK / OK > dx (1:4.4.4-7) OK > etsf-io (1.0.3-4 / 1.0.4-1~exp1) OK / OK [+] > exodusii (6.02.dfsg.1-5 / 6.02.dfsg.1-6) OK / OK > ferret-vis (6.9.3-1 / 6.9.3-2~exp1) FTBFS / OK > gdal (1.10.1+dfsg-9 / 1.11.2+dfsg-1~exp4) OK / OK > gnudatalanguage (0.9.5-2) OK > grace (1:5.1.25-1) OK > grads (2:2.0.2-5 / 2:2.0.2-6) OK / OK > gri (2.12.23-8) FTBFS [#790526] > kst (2.0.3-4) OK > libpdl-netcdf-perl (4.20-1) OK [+] > magics++ (2.24.7-3 / 2.24.7-4) FTBFS / OK > minc (2.2.00-6) OK [+] > nco (4.5.1-1) OK > ncview (2.1.5+ds1-2) OK > netcdf4-python (1.1.9-1 / 1.1.9-2~exp2) OK / OK > oasis3 (3.mct+dfsg.121022-3 / 3.mct+dfsg.121022-5) OK / OK [+] > octave-octcdf (1.1.8-1) OK > ovito (2.3.3+dfsg1-1) OK > python-scientific (2.9.4-3) OK > r-cran-ncdf4 (1.13-1) OK > r-cran-rnetcdf (1.6.3-1-1) OK > ruby-netcdf (0.7.1.1-2) OK > v-sim (3.7.2-1) OK > vtk (5.8.0-17.5 / 5.10.1+dfsg-1) OK / OK [+] > > cdo (1.6.6+dfsg.1-3 / 1.6.6+dfsg.1-4) OK / OK > gmt (5.1.2+dfsg1-1 / 5.1.2+dfsg1-2~exp1) TODO / TODO > metview (4.5.6-3 / 4.5.6-4exp1) TODO / TODO [+] > ncl (6.3.0-3~exp1 / 6.3.0-3) TODO / TODO > vtk6 (6.2.0+dfsg1-2) TODO [+] > > > Kind Regards, > > Bas > > -- > GPG Key ID: 4096R/6750F10AE88D4AF1 > Fingerprint: 8182 DE41 7056 408D 6146 50D1 6750 F10A E88D 4AF1 > > > -- > To UNSUBSCRIBE, email to [email protected] > with a subject of "unsubscribe". Trouble? Contact > [email protected] > Archive: https://lists.debian.org/[email protected] > >
